Influence of Surface-Initiated Polymerization Rate and Initiator Density

Mar 18, 2009 - Macromolecules 2017 50 (17), 6482-6488. Abstract | Full Text ... and Jiahai Zhou. ACS Applied Materials & Interfaces 2010 2 (11), 3223-...
1 downloads 0 Views 393KB Size
Macromolecules 2009, 42, 2863-2872

2863

Influence of Surface-Initiated Polymerization Rate and Initiator Density on the Properties of Polymer Brushes Hong Liu,† Min Li,‡ Zhong-Yuan Lu,*,† Zuo-Guang Zhang,‡ and Chia-Chung Sun† Institute of Theoretical Chemistry, State Key Laboratory of Theoretical and Computational Chemistry, Jilin UniVersity, Changchun 130023, China, and Key Laboratory of Aerospace Materials and Performance (Ministry of Education), School of Materials Science and Engineering, Beihang UniVersity, Beijing 100083, China ReceiVed December 18, 2008; ReVised Manuscript ReceiVed February 17, 2009

ABSTRACT: The surface-initiated polymerization with different initiator densities and different polymerization rates is investigated using molecular dynamics simulation method. We find that the initiator density, together with the polymerization rate, greatly determines the polymer brush structure, the initiation efficiency, and the graft density, especially when the initiator density is high. The excluded volume effect also plays a crucial role in the system when the chains are densely grafted. By tuning the initiator density and modifying the polymerization rate, we can obtain the polymer brushes with different degrees of polydispersity. This study partially emphasizes the importance of considering the effects of polymerization rate in further investigations.

1. Introduction Polymer brushes are generally defined as layers of polymer chains end-grafted to a surface. They give rise to a wide range of important technological and industrial applications, such as lubrications, oil recovery, colloid stability, modification of surfaces, adhesion, reversible tuning of wetting, biotechnology, and contact formation.1-5 A number of polymerization mechanisms are provided for the preparation of polymer brushes,6 such as freeradical,7 cationic,8 ring-opening metathesis polymerization,9 atom-transfer free-radical polymerization (ATRP),10-14 polymerization using 2,2,6,6-tetramethyl-1-piperidyloxy,15 anionic polymerization,16 and reversible addition-fragmentation and transfer polymerization.17,18 In recent years, a large amount of research4,5,10,11,19-46 has been carried out that focuses on polymer brushes with different graft densities. Alexander19 and de Gennes20 had theoretically demonstrated that the polymer chains tend to be strongly stretched along the direction perpendicular to the graft substrate when the graft density is high. Other researchers also pointed out that the polymer brushes will exhibit completely different conformations in solvent as the graft density is varied.47,48 In the case of low graft densities, the “mushroom” chain conformation similar to that of the free chains will appear for the endtethered chains. With increasing graft density, the chains are stretched to form the “forest” of polymer brushes because the repulsive interactions will drastically increase with increasing graft density.49 It is also important to study the effects of polymerization rate on the properties of polymer brushes. Prucker and Ru¨he50 experimentally illustrated that the activation of initiator and the diffusion of monomers to the reactive sites are the primary factors that affect the surface-initiated polymerization. They pointed out that by decreasing the probability of deactivation of the polymer radicals in ATRP, it is possible to control the layer thickness; also, if the polymerization conditions are altered in such a way, higher molecular weights will be obtained. Wittmer et al.51 indicated in their theoretical research that the brush chains are strongly extended from the surface but * Corresponding author. E-mail: [email protected]. Tel: 0086 431 88498017. Fax: 0086 431 88498026. † Jilin University. ‡ Beihang University.

extremely polydisperse because the longer chains will grow up from the layer of compact short chains to catch the outer abundant free monomers. A basic assumption in their research is that many simultaneously growing chains compete for the small influx of monomers to the surface. However, simulation studies of such systems scarcely covered the effects of polymerization rate. Matyjaszewski et al.52 and Milchev et al.53 carried out simulation studies focusing on the chain length and the graft density distributions of surfaceinitiated polymerization taking into account the living chains with activation/deactivation processes and without termination. The simulation results of Matyjaszewski et al.52 suggested broader molecular weight and chain end distributions of the polymer brushes. Milchev et al.53 pointed out that when the initiator density is high, the molecular weight distribution in the equilibrium polymer brushes obeys an N-7/4 power law. Moreover, Xiao and Wirths54 and Kim et al.55 had extended kinetic simulations with activation/deactivation equilibrium and termination in the model, and they predicted the graft mount for surface-initiated ATRP as the time elapses. Regarding the effects of polymerization rate (and most importantly, its interplay with the initiator density) on the properties of the polymer brushes, here we present a new simulation study on the surface-initiated polymerization process by utilizing a reasonably physical model similar to that in ref 53 but taking into consideration the variation of polymerization rates in the simulations. We find that the properties of polymer brushes are greatly dependent on the coupling between the initiator density and the initiation efficiency, which, however, is related to the former and the polymerization rate to a large extent. Here the initiation efficiency is defined as the ratio of the number of saturated initiators to the total number of initiators. However, the exclude volume effect plays a crucial role in such system, especially when the chains are densely grafted. As a result, in the system with higher initiator density and lower polymerization rate, the grafted chains tend to choose the conformation that greatly stretched in the direction perpendicular to the surface. By tuning the initiator density and choosing different polymerization rates, we can obtain the polymer brushes with different degrees of polydispersity. The article is organized as follows: Section 2 exhibits the simulation details, Section 3 shows the results and the corre-

10.1021/ma802817r CCC: $40.75  2009 American Chemical Society Published on Web 03/18/2009

2864 Liu et al.

Macromolecules, Vol. 42, No. 7, 2009

sponding discussion, and Section 4 presents the concluding remarks. 2. Model and Simulation Details 2.1. Truncated and Shifted Lennard-Jones Potential and Dissipative Particle Dynamics Thermostat. In this study, we consider the truncated and shifted Lennard-Jones (LJ) potential between the particles (also known as Weeks-ChandlerAndersen potential),56 in which the potential vanishes at the cutoff radius (rc ) 21/6) so that only its repulsive part is left U(r) ) ULJ(r) - ULJ(rc)

(1)

with

[( σr ) - ( σr ) ]

ULJ(r) ) 4

12

6

(2)

For simplicity, we define the units in the simulations by setting m ) 1, σ ) 1, and  ) 1. Polymer chains are constructed by connecting the neighboring monomers together via finite extensible nonlinear elastic (FENE) spring potential57,58 UFENE )

{

1 2 ln[1 - (r/l )2], (r < l ) - klmax max max

2

∞,

(3)

(r g lmax)

where the spring constant is k ) 10/σ2, a little weaker than the value in ref 5, so that the influence of the pressure variation on the system caused by bonding chain segments is slight. Here r ) | ri - rj| is the distance between neighboring monomers. The longest stretch of the bonds lmax is chosen according to the condition of bond uncrossibility59,60 that 21/2rmin > lmax, where we obtain the exact minimal distance between two particles rmin from radial distribution function (RDF),60 and then the choice of lmax is clear. The dissipative particle dynamics (DPD) thermostat is employed here to control the constant temperature. In addition to the forces originated from LJ and FENE potentials, interparticle interactions also include pairwise dissipative and random forces. They are given by61 FDij ) -γωD(rij)(vij · eij)eij and FijR ) σωR(rij)ξij∆t-1/2eij, where rij ) ri - rj, rij ) | rij|, eij ) rij/rij, and vij ) vi - vj. ξij is a random number with zero mean and unit variance. The noise level is taken to be σ ) 3 according to ref 61. The weight functions ωD(rij) and ωR(rij) of the dissipative and random forces couple together to form a thermostat. Espan˜ol and Warren62 showed the correct relations between the two functions, ωD(r) ) [ωR(r)]2 and σ2 ) 2γkBT. Following ref 63, we take the form of ωD(r) and ωR(r) to be ωD(r) ) [ωR(r)]2 )

{

[1 - (r/rc)]2 (r < rc) (r g rc) 0

(PBC) are applied in both x and y directions, but in the z direction, two layers of regularly arranged and densely packed frozen particles are utilized to form the wall. The interaction parameters between particles are set to be the same; that is, MM ) MW ) WW ) 1, in which MM denotes the repulsive strength between two monomers, MW between a monomer and a wall particle and WW between two wall particles. Some of the wall particles in the upper layer of the bottom wall are chosen to be the initiators so that the polymerization starts from the surface of the bottom wall to generate the end-tethered chains. The distance between the layers in a wall, dL, is determined by the density of the wall particles, FW; that is, dL ) FW-1/3. (FW is a preset parameter, denoting the number of wall particles in unity volume, so dL is the distance of the neighboring wall particles in the x, y, and z directions.) Because the interaction cutoff radius rc ) 21/6, the monomers are not able to feel the existence of the wall particles from the third layer, even if a finitely larger FW (for example, FW ) 7) is chosen. Therefore, the impenetrable wall can be reasonably described by two closely packed layers of frozen wall particles. In the simulations, the temperature is set equal to ; that is, kBT )  ) 1. The particle density is kept as F ) 0.85.63 According to ref 63, the Schmidt number, Sc, in such a system is roughly Sc ≈ 30, corresponding to a value for a real fluid. The reduced box size is Lx ) Ly ) 16.837, and Lz ) 56.123. (The brushes are grafted from the bottom wall at Z ) -Lz/2 ) -28.0615, and the top wall is at Z ) Lz/2 ) 28.0615.) Therefore, the system consists of 13 523 particles. Here we choose a larger Lz compared with Lx and Ly so that there are relatively larger spaces for the polymer brushes to grow up, and thus we can observe the processes of fast polymerization even in a relatively long period. 2.3. Modeling the Polymerization. In the bottom wall, some orderly distributed initiating sites are selected that represent an initiator density σi, defined as σi ) Ni/(Lx × Ly), where Ni denotes the number of initiators. Here we employ the idea in our previous model64 by introducing the polymerization probability, Pr, to control the process of polymerization. In each reaction interval, τ, if an active end meets several monomers in the interaction radius, then it first randomly chooses one of the monomers as an reacting object. Subsequently, another random number is generated, and by checking if it is smaller than the preset Pr, we decide whether the monomer will connect with the active end. During polymerization, the newly connected monomers then turn to be the growth centers in the next propagation step of the same chain to connect other free monomers so that the active end is transferred forward. In this model, the chain termination step is omitted, which is the same as that in ref 65. It is easy to find that there is a specific relationship between the polymerization rate, rp, and the reaction probability, Pr

(4)

DPD thermostat acting on an LJ system was evaluated as a potentially useful technique for controlling the temperature.5,63 In ref 63, the authors chose a time step ∆t ) 0.01, and in ref 5, the authors chose smaller values: ∆t ) 0.0005 or 0.002. Here because of the existence of the walls in the simulations, we have to pursue a balance between the influence of heat/ momentum dissipation caused by the walls on the temperature fluctuation and the control of temperature by the DPD thermostat. We thus find that ∆t ) 0.005 is a reasonable value in our simulation. 2.2. Model Construction. In this study, the model used to describe the surface-initiated polymerization is composed of free monomers and planar walls. The periodic boundary conditions

rp ) -

[P*]Pr d[M] ) dt τ

(5)

where [M] is the free monomer concentration, [P*] is the concentration of growth centers, and τ is the reaction time interval. Therefore, the model corresponds to the polymerization process with a constant reaction rate. In the simulations, we choose five typical values of Pr (Pr ) 0.0005, 0.001, 0.002, 0.005, and 0.01) to study the effect of the polymerization rate on the properties of the polymer brushes. We assume that the different polymerization probabilities correspond to different polymerization rates of the systems determined by different levels of reaction activation energies.64 Here Pr ) 0.01 corresponds to a fast process of polymerization, and Pr ) 0.0005 corresponds to a slow one. Because the polymerization can be

Macromolecules, Vol. 42, No. 7, 2009

Figure 1. Snapshots of the surface-initiated polymerization with (a) a low initiator density σi ) 0.032 and (b) a high initiator density σi ) 0.508. The densely packed green beads represent the wall particles, the purple beads represent the initiators, the gray beads represent the grafted monomers, and the red beads represent the active ends of the chains. The free monomers are omitted here for clarity.

controlled under relatively low reaction rate in our model, it is not necessary to specially add the activation/deactivation process as in ATRP for the active ends. We consider the reversible activation process to be the fine details on the atomistic level inside the active ends of the chains, which can be reasonably ignored in our coarse-grained model. Although the general concept of living polymerization in the soft matter physics community often means reversible polymerization, here we reasonably ignore the concept difference and consider our model to be a kind of living because it is controllable under low reaction rate. This handling causes our model to be a little different from that of ref 53, in which the authors considered the reversible polymerization; that is, the end bonds of the chains have the probability of breaking, which is associated with a true Hamiltonian. Therefore, in that model, the system is in full equilibrium, both for the generalized coordinates and for the topology. Nevertheless, in our model, the chemical bonds of the chains correspond to essentially infinite energy gain. Therefore, it is a practically irreversible polymerization model. Different behavior should be expected for irreversible polymerization and reversible polymerization, as in ref 53. In the irreversible polymerization, the mass distribution is not at thermal equilibrium but is imposed by the dynamical pathway. Because of this, it can be hoped that very different brush structures might be synthesized. Figure 1a shows the snapshot of the surface-initiated polymer chains with a low initiator density, σi ) 0.032, in which only nine chains are grown up. The snapshot is obtained by extracting the grafted particle coordinates just a few time steps after the polymerization is originated so that the lengths of the chains are short. (The free monomers are eliminated in the Figure for clarity.) In comparison, Figure 1b exhibits the snapshot of the grown-up short chains with a higher initiator density σi ) 0.508. 3. Results and Discussion The simulations are conducted in canonical ensemble. The Groot-and-Warren-modified velocity Verlet algorithm is used

Surface-Initiated Polymerization 2865

Figure 2. Time evolution of the distance of the center of mass of the grafted chains from the grafting plane (Zcm) (a) with different σi at Pr) 0.001 and (b) with σi ) 0.508 but different Pr. For each data point, four independent samples are simulated under the same condition to calculate Zcm mean value and the error bar.

for numerical integration.61 The initial configuration is set to be the randomly distributed free monomers in the system, then 1 × 105 steps molecular dynamics (MD) integration is carried out to relax the system. Consequently, a process of polymerization with 1 × 106 time steps is modeled. Here we choose the value of reaction time interval to be τ ) 0.05; that is, the polymerization of a specific active end may take place every 10 time steps depending on Pr because ∆t ) 0.005. 3.1. Controllability of Polymerization Process. A significant characteristic of living polymerization is that it is controllable; that is, during the polymerization process, the distance between the center of mass of the grafted chains and the grafting plane should increase linearly in the substrate normal direction. Therefore, we have checked our surface-initiated polymerization model on this issue. Figure 2a shows the time evolution of the distance between the center of mass of the grafted chains and the plane, Zcm, with different σi at Pr ) 0.001, which corresponds to a relatively slow polymerization rate (Here, Zcm can also be considered to be the mean layer half-height of the film). It is clearly shown in Figure 2a that for the systems with different initiator densities, Zcm exhibits an almost linear increase. The system with larger initiator density yields a larger increasing slope. In the case of such a slow reaction rate (Pr ) 0.001), the competitions of the propagation between different chains are very weak. In other words, the ambient free monomers are abundant for the chains to propagate. However, it is reasonable to imagine that the growth behaviors of the systems with the same reaction rate but different initiator densities are different. In the case of low initiator densities, the tethered chain configuration is “mushroom”-like, whereas densely grafted chains will be obliged to grow nearly straight and form the forest of brushes. Therefore, the polymer brushes with larger values of σi exhibit higher increasing slope between Zcm and time. However, when we consider a faster reaction rate, the model polymerization will become slightly uncontrollable. Figure 2b

2866 Liu et al.

shows the time evolution of Zcm of the grafted chains with σi ) 0.508 but different Pr. It is obvious that in the systems with faster polymerization rates, the curves of Zcm ∼ t deviate from the linear style more or less, showing the uncontrollable feature. The Zcm ∼ t curve for Pr ) 0.01 reaches a plateau with Zcm ≈ 28.0615 in the late stage of polymerization, implying that the center of mass of the polymer brushes is just at the middle between the two walls, namely, the brushes have completely filled in the simulation box. The slightly nonlinear part of each curve in the early stage is reasonable, during which the film of the polymer brushes is not yet formed, and only very short chains exist. Therefore, it is necessary to choose slow-tomoderate reaction rates in the model to keep it a controllable living polymerization and at the same time to guarantee the efficiency of polymerization. Therefore, for comparison in the following simulations, Pr ) 0.001, 0.002, and 0.005 are chosen as the reaction probabilities of living polymerization to represent relatively slow, moderate, and fast reaction rates, respectively. It should be specially stressed that Zcm ∼ t does not increase linearly ideally, implying complex influences from possible factors such as the diffusion limited aggregation (DLA),66,67 the relaxation of the brush chains, the screening effect of the grown chains on the unsaturated initiators, and so on. In the systems, only part of the free monomers participate in the polymerization, the relationship Zcm ∼ t is expected to be dominantly governed by DLA. 3.2. Influence of Wall Roughness. In this model, we construct a wall by regularly arranging the densely packed and fixed wall particles in two planar layers. Therefore, it is not the type of idealized flat wall and the wall roughness is greatly dependent on the density of the wall particles, FW. When we choose a larger value of FW, we can obtain a smoother wall because the wall particles are more densely packed. However, it is simply unphysical if the neighboring particles are too close. However, to model different initiator densities, it is necessary to change the value FW so that we can finely tune the number of wall particles to make the value σi reasonably represent the global initiator density of the surface without any local differences. Therefore, studying the influence of the wall roughness on the system is actually important. As a kind of neutral wall, its influence is only restricted to the polymer brushes or free monomers near the wall; that is, the walls with different FW values may provide different repulsive strengths to the end-tethered polymer brushes so that the latter may exhibit different stretching degrees in the z direction only in a layer near the wall. Therefore, we use the second Legendre polynomial P2(cos θ1) ) [3〈cos2 θ1〉 - 1]/2 as a measure for checking the averaged first bond orientation of the polymer brushes,5,53,68 which can be taken to be a criterion for the repulsive strength from the wall. Here θ1 is defined as the angle between the first bond vector (connecting the anchoring initiator and the first grafted monomer) and the z axis. Therefore, P2 ) -0.5 denotes the bond orientation of lying parallel to the wall, P2 ) 1 represents the bonds lying perpendicular to the wall, and P2 ) 0 represents the bonds that are randomly oriented. Figure 3 shows the averaged first bond orientation of the polymer brushes with different FW values. It is clearly shown in Figure 3 that when FW ) 1, the wall is too rough and cannot provide strong repulsion for the first grafted monomers of the polymer brushes. As a result, the averaged value of P2 of the first bond is 1 denotes the fact that the chain generally stretches along the z direction. The value of φs indicates the degree of the chain stretching. Figure 9a shows the time evolution of φs in the cases of Pr ) 0.001 but different σi. Here we choose the small polymerization rate, Pr ) 0.001, to gain the high initiation efficiency. For the systems with low initiator densities, the value of φs is near 1, indicating that the chain configuration is roughly a spherical coil; that is, the polymer brushes possess mushroom configuration. φs increases

2870 Liu et al.

Figure 10. Normalized average active end density distribution along the z direction in the case of σi ) 0.127 and 0.903 but different Pr. The red sets denote the systems with σi ) 0.127, and the black sets denote σi ) 0.903. Five independent samples, 1000 snapshots obtained in each, are simulated under the same condition to calculate the mean value and the error bar. The error bars are not shown here for clarity.

with increasing σi, denoting that the chains are stretched when the local environment is crowded. The systems with very high initiator densities (e.g., σi ) 0.691 and 0.903) yield almost the same values of φs, which shows that in the two systems the local environments for the chains are similar and the chains cannot become more crowded. In almost all of the systems, the value of φs increases slightly as time elapses. This phenomenon suggests that during the chain growth process, Rg increases, and thus the excluded volume effect may further induce the stretch along the z direction. In Figure 9b we present the comparison of φs for the systems with the same σi but different Pr. In the case of low initiator density with σi ) 0.032, we can find that the three systems with different Pr values possess the similar stretch parameter, that is, φs ≈ 1. The system with Pr ) 0.005 has a slightly larger φs in the late stage just because the chains grow faster in this system. However, in the case of high initiator density with σi ) 0.903, we find apparent differences between systems with different Pr values. The system with Pr ) 0.001 has the larger value of φs, and the system with Pr ) 0.005 has the smaller value. This is simply because in the case of high initiator density σi ) 0.903, the system with Pr ) 0.001 has a higher initiation efficiency such that the chains are too crowded and have to stretch upward. On the contrary, the systems with faster polymerization rates have lower initiation efficiency, and the chains have enough space to diffuse laterally. 3.7. Active End Distribution. To investigate the active end distribution in the cases of different σi and Pr further, in Figure 10, we show the normalized average active end density distribution along the z direction Fe(Z) (defined as the active end density, Fe(Z), divided by the number of initiators, Ni; that is, Fje(Z) ) Fe(Z)/Ni) with different σi and Pr. In the case of high initiator density with σi ) 0.903, one can find more active ends in the region near the surface. However, in the case of low initiator density with σi ) 0.127, the result is on the contrary. As we explained above, because of the screening effect of the grown chains on the unsaturated initiators, the initiation efficiency is always low in the systems with high σi. Therefore, there is a large fraction of unsaturated initiators in the case of σi ) 0.903 as a result of the low initiation efficiency. Therefore, we find a fairly concentrated distribution of active ends near the surface when σi ) 0.903. In comparison, σi ) 0.127 is lower than the critical value of initiator density, as presented in Figure

Macromolecules, Vol. 42, No. 7, 2009

Figure 11. Variation of polydispersity index (PDI) of the grafted polymer chains with increasing σi at different polymerization rates. Four independent samples are simulated under the same condition to calculate PDI mean value and the error bar.

4a, so the initiation efficiency is high. Almost all of the initiators can grow up into chains in the systems with σi ) 0.127, and thus the active ends can distribute everywhere along the z direction. This result is completely different from the simulation results with monodisperse polymer brush model, for example, figure 4 in ref 2, in which they obtained a near-Gaussian active end distribution. This is because they have not taken the influence of the polymerization process into consideration. 3.8. Polydispersity and Mass Distribution. A very important parameter for the growth of polymer brushes is the polydispersity index (PDI), where PDI )

jw M ) jn M

∑ NM /∑ NM ∑ NM/∑ N 2 i

i

i

i

i

i

i

i

i

i

(9)

i

We present the mean values of PDI for each system with different σi and Pr in Figure 11. At the same Pr, PDI of the grafted chains increases with increasing initiator density. When σi is very low, PDI is a little larger than 1.0, indicating a homogeneous growth process of the chains. When σi increases to larger values, PDI increases sharply, implying the emergence of the screening effect of the grown chains, which leads to the contrast of the growth conditions of short and long chains. At the same initiator density, the systems with faster polymerization possess a larger PDI, which supports the fact that a serious screening effect of the grown chains occurs in such systems. If σi and Pr are simultaneously high, then a large part of the initiators have to remain unsaturated (N ) 1) or to grow to very short chains. Therefore, it can be expected to obtain a very large value of PDI in such systems. According to the results concerned, if one needs to keep the polydispersity of high density brushes as low as possible, then the surface polymerization rate should be kept as slow as possible. A relatively homogeneous growth condition of the chains would profit the occurrence of low dispersity. Here we further compare our results with available experimental data. Devaux et al.70 had achieved very high graft density (1.10 chain/nm2) of the polystyrene chains grown on the silicon surface at fairly low polydispersity (PDI ≈ 1.1). According to the molar mass (104.15 g/mol) and the density (0.909 g/cm3) of styrene, we can roughly calculate that the diameter of a “coarse-grained” styrene bead is 7.14 Å. If we normalize the diameter to be 1, then we can find that the reduced graft density

Macromolecules, Vol. 42, No. 7, 2009

Surface-Initiated Polymerization 2871

Figure 12. Mass distribution p(N, t) at different times after the start of polymerization in the systems with different σi and Pr.

of 1.10 chain/nm2 should be 0.56 in reduced unit in our model. Because the unsaturated initiators are considered to be the chains with length N ) 1 in the calculation of PDI, here the initiator density, σi, and the so-called “graft density” could be considered to be equivalent. (Note that the concept “graft density” in ref 70 means the density of the really initiated chains in unity area, which is not the same as that in our article.) Therefore, we can have a reasonable comparison between the experimental results and the current simulation. From Figure 11, we find that the value should be PDI ) 1.27 at σi ) 0.56 when Pr ) 0.0005, which is a little larger than the experimental result (PDI ) 1.10). Because faster reaction rates correspond to larger values of PDI, if we further decrease the polymerization rate, then a PDI value close to 1.10 can be expected to be obtained. Figure 12 shows the mass distribution p(N, t) at different times after the start of polymerization in the systems with different σi and Pr. In the case of low initiator density and slow reaction rate (Figure 12a), we can find the mass distribution to be Poisson-like, implying the competition-absent growth like the free chains. When the initiator density is low but the reaction rate is fast (Figure 12b), p(N, t) is still Poisson-like in the beginning of the polymerization. However, the fast chain growth leads to screening effect and great chain length difference. Therefore, the Poisson-like shape of p(N, t) is replaced by a much broader distribution. In the case of high initiator density (Figure 12c,d), the initiation efficiency is the dominant factor, showing an extremely high distribution of the mass near N ) 1 (N ) 1 corresponds to the unsaturated initiators). It should be noted that σi ) 0.903 corresponds to the limit condition compared with experiments. The very high graft density in ref 70 of 1.10 chain/nm2 corresponds to 0.56 in our model. We find that the first value of p(N, t) in Figure 12c is smaller than

that in Figure 12d, which implies fewer unsaturated initiators in this system, that is, higher initiation efficiency when Pr is small. This conclusion agrees with that in Section 3.5. Furthermore, we find no difference between the first values of p(N, t) in Figure 12d at different times, which indicates the strong screening effect of the fast grown chains. Figure 12c,d has low distribution at the region of large chain lengths. This result can be easily understood. In the system with high σi, the small part of surviving chains from the competition of the growth near the grafting plane obviously has less hindrance to grow longer and longer. Therefore, we can find a broad distribution of the chains as the chain length increases. 4. Conclusions In this study, the surface-initiated polymerization with different initiator densities and polymerization rates has been investigated using MD simulations. By characterizing the time evolution of the distance between the center of mass of the grafted monomers and the grafting plane, we find that it is necessary to choose slow-to-moderate reaction rates in the model to keep it a controllable living polymerization. We then study the variation of the grafted monomer fraction, fG, with increasing σi at different polymerization rates. In agreement with the experimental results, we obtain a window of σi as the critical value of the initiation efficiency. When the initiator density is lower than the critical value, the initiation efficiency is nearly 100%. In particular, the graft density profile in the z direction, whose variation reflects the growth of the polymer film, is studied. By comparing the absolute and normalized graft densities along the z direction of different systems, we find that when initiator density is low, the initiation efficiency is always

2872 Liu et al.

high. However, in the systems with high initiator density, the slow or moderate polymerization yields relatively higher initiation efficiency. Moreover, the time evolution of the fraction of unsaturated initiators is characterized. We find a screening effect of the grown chains on the unsaturated initiators when the initiator density is high, which may be one of the main reasons that causes universal low initiation efficiency in the systems with high initiator density. We have also studied the bond orientation function P2(cos θn) with the bond index, n, and the time evolution of the chain stretch parameter, φs. We find that the conformation of the chains mainly depends on the excluded volume effect. Therefore, the systems with dilute chains commonly possess mushroom configuration, and the systems with densely grafted chains are often obliged to stretch along the surface normal direction. The -0.49 scaling between P2(cos θn) and t, characterizing the excluded volume effect, is found also dependent on the polymerization rate. The normalized active end density distribution is also investigated, and broader chain end distribution is obtained than that in the monodisperse polymer brushes. Besides, the observed fairly concentrated distribution of unsaturated initiators near the surface again implies the screening effect of the grown chains on the unsaturated initiators. Moreover, the influences of σi and Pr on the polydispersity as well as the mass distribution of the grown chains are especially investigated. A relatively slow polymerization and low initiator density will profit the occurrence of low dispersity and narrow mass distribution. In this study, we find that the properties of the polymer brushes are greatly dependent on the coupling between the initiator density and the polymerization rate. The excluded volume effect also plays a crucial role in the system, especially when the chains are densely grafted. By tuning the initiator density and modifying the polymerization rate, we can obtain the polymer brushes with different degrees of polydispersity. This study partially emphasizes the importance of considering the effects of polymerization rate in further investigations. Acknowledgment. This work is supported by the National Science Foundation of China (20774036, 50803002) and Fok Ying Tung Education Foundation (114018). References and Notes (1) Pal, S.; Seidel, C. Macromol. Theory Simul. 2006, 15, 668. (2) Chen, C.-M.; Fwu, Y.-A. Phys. ReV. E 2000, 63, 011506. (3) Goujon, F.; Malfreyt, P.; Tildesley, D. J. Chem. Phys. Chem. 2004, 5, 457. (4) Malfreyt, P.; Tildesely, D. J. Langmuir 2000, 16, 4732. (5) Pastorino, C.; Binder, K.; Kreer, T.; Mller, M. J. Chem. Phys. 2006, 124, 064902. (6) Advincula, R. AdV. Polym. Sci. 2006, 197, 107. (7) Prucker, O.; Ru¨he, J. Macromolecules 1998, 31, 592. (8) Jordan, R.; Ulman, A. J. Am. Chem. Soc. 1998, 120, 243. (9) Weck, M.; Jackiw, J. J.; Rossi, R. R.; Weiss, P. S.; Grubbs, R. H. J. Am. Chem. Soc. 1999, 121, 4088. (10) Yamamoto, S.; Tsujii, Y.; Fukuda, T. Macromolecules 2000, 33, 5995. (11) Jones, D. M.; Brown, A. A.; Huck, W. T. S. Langmuir 2002, 18, 1265. (12) Ejaz, M.; Yamamoto, S.; Ohno, K.; Tsujii, Y.; Fukuda, T. Macromolecules 1998, 31, 5934. (13) Kong, X.-X.; Kawai, T.; Abe, J.; Iyoda, T. Macromolecules 2001, 34, 1837. (14) von Werne, T.; Patten, T. E. J. Am. Chem. Soc. 2001, 123, 7497. (15) Husseman, M.; Malmstrm, E. E.; McNamara, M.; Mate, M.; Mecerreyes, D.; Benoit, D. G.; Hedrick, J. L.; Mansky, P.; Huang, E.; Russell, T. P.; Hawker, C. J. Macromolecules 1999, 32, 1424. (16) Zhou, Q.; Nakamura, Y.; Inaoka, S.; Park, M.; Wang, Y.; Mays, J.; Advincula, R. PMSE [Prepr.] 2000, 82, 291. (17) Baum, M.; Brittain, W. J. Macromolecules 2002, 35, 610. (18) Sedjo, R. A.; Mirous, B. K.; Brittain, W. J. Macromolecules 2000, 33, 1492. (19) Alexander, S. J. Phys. (Paris) 1977, 38, 983.

Macromolecules, Vol. 42, No. 7, 2009 (20) de Gennes, P. G. Macromolecules 1980, 13, 1069. (21) de Gennes, P. G. AdV. Colloid Interface Sci. 1987, 27, 189. (22) Milner, S. T.; Witten, T. A.; Cates, M. E. Macromolecules 1988, 21, 2610. (23) Milner, S. T. Science 1991, 251, 905. (24) Kuznetsov, D. V.; Chen, Z.-Y. J. Chem. Phys. 1998, 109, 7017. (25) Irfachsyad, D.; Tildesley, D.; Malfreyt, P. Phys. Chem. Chem. Phys. 2002, 4, 3008. (26) Wijmans, C. M.; Smit, B. Macromolecules 2002, 35, 7138. (27) Murat, M.; Grest, G. S. Phys. ReV. Lett. 1989, 63, 1074. (28) Dickman, R.; Anderson, P. E. J. Chem. Phys. 1993, 99, 3112. (29) Lai, P.-Y.; Zhulina, E. B. Macromolecules 1992, 25, 5201. (30) Lai, P.-Y.; Binder, K. J. Chem. Phys. 1993, 98, 2366. (31) Lai, P.-Y.; Binder, K. J. Chem. Phys. 1991, 95, 9288. (32) Lai, P.-Y.; Binder, K. J. Chem. Phys. 1992, 97, 586. (33) Grest, G. S. J. Chem. Phys. 1996, 105, 5532. (34) Doyle, P. S.; Shaqfeh, E. S. G.; Gast, A. P. Macromolecules 1998, 31, 5474. (35) Daoulas, K. C.; Terzis, A. F.; Mavrantzas, V. G. J. Chem. Phys. 2002, 116, 11028. (36) Laradji, M.; Guo, H.; Zuckermann, M. J. Phys. ReV. E 1994, 49, 3199. (37) Karim, A.; Satija, S. K.; Douglas, J. F.; Ankner, J. F.; Fetters, L. J. Phys. ReV. Lett. 1994, 73, 3407. (38) Hu, T.-J.; Wu, C. Phys. ReV. Lett. 1999, 83, 4105. (39) Hadziioannou, G.; Patel, S.; Granick, S.; Tirrell, M. J. Am. Chem. Soc. 1986, 108, 2869. (40) Taunton, H. J.; Toprakcioglu, C.; Fetters, L. J.; Klein, J. Nature 1988, 332, 712. (41) Taunton, H. J.; Toprakcioglu, C.; Fetters, L. J.; Klein, J. Macromolecules 1990, 23, 571. (42) Cosgrove, T. J. Chem. Soc., Faraday Trans. 1990, 86, 1323. (43) Auroy, P.; Auvray, L.; Le´ger, L. Phys. ReV. Lett. 1991, 66, 719. (44) Patel, S.; Tirrell, M. Annu. ReV. Phys. Chem. 1989, 40, 597. (45) Yamamoto, S.; Ejaz, M.; Tsujii, Y.; Fukuda, T. Macromolecules 2000, 33, 5608. (46) Ma, H.-W.; Hyun, J.-H.; Stiller, P.; Chilkoti, A. AdV. Mater. 2004, 16, 338. (47) Halperin, A.; Tirrell, M.; Lodge, T. P. AdV. Polym. Sci. 1992, 100, 31. (48) Kawaguchi, M.; Takahashi, A. AdV. Colloid Interface Sci. 1992, 37, 219. (49) Lai, P.-Y.; Halperin, A. Macromolecules 1991, 24, 4981. (50) Prucker, O.; Ru¨he, J. Langmuir 1998, 14, 6893. (51) Wittmer, J. P.; Cates, M. E.; Johner, A.; Turner, M. E. Europhys. Lett. 1996, 33, 397. (52) Matyjaszewski, K.; Miller, P. J.; Shukla, N.; Immaraporn, B.; Gelman, A.; Luokala, B. B.; Siclovan, T. M.; Kickelbick, G.; Vallant, T.; Hoffmann, H.; Pakula, T. Macromolecules 1999, 32, 8716. (53) Milchev, A.; Wittmer, J. P.; Landau, D. P. J. Chem. Phys. 2000, 112, 1606. (54) Xiao, D.-Q.; Wirth, M. J. Macromolecules 2002, 35, 2919. (55) Kim, J. B.; Huang, W.-X.; Miller, M. D.; Baker, G. L.; Bruening, M. L. J. Polym. Sci., Part A: Polym. Chem. 2003, 41, 386. (56) Weeks, J. D.; Chandler, D.; Andersen, H. C. J. Chem. Phys. 1971, 54, 5237. (57) Bird, R. B.; Armstrong, R. C.; Hassager, O. Dynamics of Polymeric Liquids; Wiley: New York, 1977; Vols. 1 and 2. (58) Kremer, K.; Grest, G. S. J. Chem. Phys. 1990, 92, 5057. (59) Nikunen, P.; Vattulainen, I.; Karttunen, M. Phys. ReV. E 2007, 75, 036713. (60) Liu, H.; Xue, Y.-H.; Qian, H.-J.; Lu, Z.-Y.; Sun, C.-C. J. Chem. Phys. 2008, 129, 024902. (61) Groot, R. D.; Warren, P. B. J. Chem. Phys. 1997, 107, 4423. (62) Espan˜ol, P.; Warren, P. B. Europhys. Lett. 1995, 30, 191. (63) Soddemann, T.; Dnweg, B.; Kremer, K. Phys. ReV. E 2003, 68, 046702. (64) Liu, H.; Qian, H.-J.; Zhao, Y.; Lu, Z.-Y. J. Chem. Phys. 2007, 127, 144903. (65) Akkermans, R. L. C.; Toxvaerd, S.; Briels, W. J. J. Chem. Phys. 1998, 109, 2929. (66) Witten, T. A., Jr.; Sander, L. M. Phys. ReV. Lett. 1981, 47, 1400. (67) Ball, R.; Nauenberg, M.; Witten, T. A., Jr. Phys. ReV. A 1984, 29, 2017. (68) He, G.-L.; Merlitz, H.; Sommer, J.-U.; Wu, C.-X. Macromolecules 2007, 40, 6721. (69) Tsujii, Y.; Ohno, K.; Yamamoto, S.; Goto, A.; Fukuda, T. AdV. Polym. Sci. 2006, 197, 1. (70) Devaux, C.; Cousin, F.; Beyou, E.; Chapel, J.-P. Macromolecules 2005, 38, 4296.

MA802817R